Abstract
A new statistical testing approach using a weighted logrank statistic is developed for rodent tumorigenicity assays that have a single terminal sacrifice but not cause-of-death data. Instead of using cause-of-death assignment by pathologists, the number of fatal tumors is estimated by a constrained nonparametric maximum likelihood estimation method. For data lacking cause-of-death information, the Peto test is modified with estimated numbers of fatal tumors and a Fleming-Harrington-type weight, which is based on an estimated tumor survival function. A bootstrap resampling method is used to estimate the weight function. The proposed testing method with the weight adjustment appears to improve the performance in various situations of single-sacrifice animal experiments. A Monte Carlo simulation study for the proposed test is conducted to assess size and power of the test. This testing approach is illustrated using a real data set.
Original language | English (US) |
---|---|
Pages (from-to) | 95-113 |
Number of pages | 19 |
Journal | Environmental and Ecological Statistics |
Volume | 12 |
Issue number | 1 |
DOIs | |
State | Published - Mar 2005 |
Keywords
- Bioassay
- Bootstrap
- Competing risk
- Fatal tumor
- Single sacrifice
- Weight function
ASJC Scopus subject areas
- Statistics and Probability
- General Environmental Science
- Statistics, Probability and Uncertainty